PanLex: Goal


The vision underlying the PanLex project is that the world’s thousands of human languages can flourish over the long-term future. To help make them all useful for global communication, PanLex aims to translate any word (or word-like phrase such as “sweet potato”) from any language into any other language. With PanLex, it could become as easy to translate a word from Benua into Bengali as from English into French. High-quality communication across languages usually requires more than lexical translation (translating words), but lexical translation is one of its main prerequisites.

How difficult is this? Most authorities estimate that there are about 7,000 languages in the world today. Suppose they contain on average about 50,000 words each. This would yield a rough estimate of 350 million words. A system that translates any word into any other language would, then, need to provide over 2 trillion (2,000,000,000,000) translations. The vast majority of these couldn’t be found in any dictionary, so the system would need to infer them, using artificial intelligence. By documenting all the known lexical translations, the PanLex project can minimize the need for, and maximize the quality of, automatically inferred lexical translations.

To facilitate lexical translation among all languages, our immediate goal is to construct a comprehensive lexical translation database. This differs somewhat from a panlingual “dictionary”:

Widespread practical use of PanLex will depend on our success in expanding the database and on researchers and developers making use of it. At present the major users of PanLex are search engines; they query the database via PanLinx about 270 million times per year.

Valid XHTML 1.1!